In-silico screening of small compounds against Lassa fever haemorrhagic virus nucleoprotein

Abstract

Lassa fever hemorrhagic virus (LFHV) outbreaks in Nigeria continue to increase. With the possibility of a potentially deadly pandemic and the absence of a vaccine, an effective candidate drug that presents with no adverse effects and better clinical outcomes is urgently needed; hence, the aim of the study. Retrieved ligands: 5-(Methylene) evodiamine, N-(2-Fluorobenzoyl) evodiamine, N-[[2-(3-chlorophenyl)-5-(trifluoromethyl) pyrazol-3-yl]methyl]-2-(1-methyl-2,3-dihydroindol-5-yl) propanamide (NCTPMMDP), lauric acid, and sofosbuvir, and the target protein, 3MX5, were prepared and utilised for docking, molecular dynamics simulation, ADMET profiling, and DFT using standard protocols. The docking score for lauric acid was the lowest, with a value of − 6.1, followed by the standard drug and co-crystal control drug with scores of − 9.1 and − 8.9 kcal/mol, respectively. On the other hand, the scores of the other ligands ranged from − 9.5 to − 12.3 kcal/mol. Molecular simulations revealed fluctuations in the RMSD and RMSF values; however, overall, the formed complexes were stable during most of the docking simulation runs. Principal component analysis (PCA) analysis showed that all the ligands, apart from lauric acid, showed better conformational changes within the backbone of the protein. ADMET profiling showed that the ligands were better oral drug candidates than sofosbuvir, as revealed by their better absorption values and absolute compliance with the Lipinski rule of five. The DFT result showed that, apart from lauric acid, all the test ligands possess small energy gaps indicative of high reactivity. Put together, these findings indicate that the test ligands possess anti-LHFV potential that merits further in vivo and in vitro evaluations.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-89989-9.

Introduction

The Lassa fever virus was first discovered in 1969 in Lassa town in Borno State, Nigeria, hence its name¹. The virus is a member of the Arenaviridae family, and it is a single-stranded RNA virus². Since the index case, the LHFV has continued to spread across Nigeria with no end in sight³. As of April 2023, there were 4702 and 877 suspected and confirmed cases, respectively in Nigeria reported within the first quarter of 2023³. A weekly update from the National Centre for Disease Control (NCDC) for epidemiological week 7 reported a total of 507 confirmed cases, 66 suspected cases, and 13 deaths for several states, and these included Edo, Ondo, Kogi, Ebonyi, Bauchi, Cross River, Taraba, Enugu, Imo, Delta, Niger, and Yobe States in Nigeria⁴. Once limited to Nigeria, the virus has now spread to other countries in the West African region were cases have also been reported⁵.

Interestingly, in all these countries, the outbreaks appear to overlap with the prevalence of wild multimammate rats (Mastomys species)^6–8. The implicated rodent species comes into contact with food, water, and various surfaces in various homes and, in the process, leave behind their droppings. At the moment, human transmission occurs via contact of the virus from rodent droppings with compromised skin or mucous membrane⁹. Human-to-human transmission cycle has been reported to be via infected secretions that include blood, saliva, and urine⁹. Once a human infection occurs, death is imminent from organ failure within 10–14 days, during which the infected persons present with its unique symptoms⁹.

At the moment, there is no approved LFHV vaccine, but there are reports of a first-in-human trial^10,11. The candidate vaccine study reported some adverse events that were limited to mostly mild or moderate in severity and good safety profiles following its administration¹¹. Uptake of vaccines in Africa as a whole has been a problem. The acceptance of the COVID-19 vaccine in Africa was far less than that of the rest of the world^12,13. These issues put together necessitate the urgent need for an approved drug for the virus that can be used to effectively treat cases⁵. Among the numerous targets for drug development, the viral nucleoprotein is an important option, as it is involved in various aspects of the virus life cycle⁵. Earlier studies relied on the use of rational drug design in the search for a novel therapeutic agent for LHFV^14–16. Odey et al.¹⁵ reported docking scores that ranged from − 6.0 to − 7.7 kcal/mol for cissampeline against various LFHV proteins. In another study, rheosmin was reported to have a binding score of − 6.5 kcal/mol when docked against the 3mWp protein. Eyo et al.¹⁴ reported feruloyltyramine’s excellent docking activities against various LFHV proteins.

We aimed to screen four (4) small compounds against the LFHV nucleoprotein 3MX5 using in silico approaches. The compounds were selected for several reasons. First, the selection was based on their structural diversity, reported bioactivities, and potential to interact with the target protein, 3MX5. Compounds with PubChem IDs 56,967,508 and 49,804,912 are well-known evodiamine derivatives whose anti-virus properties are well known¹⁷. On the one hand, lauric acid (PubChem ID 3893), a saturated fatty acid, was selected to explore the effect of hydrophobic interactions on ligand binding. Furthermore, the compound PubChem ID 45,375,808 was selected based on its structural complexity and predicted high binding affinity, while PubChem ID 49,804,500 was selected as it provides diverse functional group diversity for exploring hydrogen bonding and van der Waals interactions. Put together, the ligands were selected to provide a detailed investigation into binding dynamics and stability as they complex with the target protein and to also provide more insight into their pharmacokinetic properties, laying the groundwork for future therapeutic developments.

Methods

Ligand and protein retrieval

The ligands utilised in this study were 5-(Methylene) evodiamine (PubChem ID 56967508), lauric acid (PubChem ID 3893), N-(2-Fluorobenzoyl) evodiamine (PubChem ID 49804912), and N-[[2-(3-chlorophenyl)-5-(trifluoromethyl) pyrazol-3-yl]methyl]-2-(1-methyl-2,3-dihydroindol-5-yl) propanamide (NCTPMMDP) (PubChem ID 49804500) and sofosbuvir (PubChem ID 45375808)^17–23. On the other hand, the target protein, 3MX5, was obtained from the Research Collaboratory for Structural Bioinformatics (RCSB) website²³. The published properties of the protein indicate that it was obtained via X-ray diffraction and had a resolution of 1.90 Å.

Preparation of the ligands

The ligands, including the controls (n = 6), were prepared as previously reported using the ChemDraw software²⁴. The energies of the ligands were minimised using default settings, their SMILES strings retrieved for ADMET prediction, and the resulting minimised structure was stored using the MDL SD file (^.sdf) format²⁵.

Preparation of the target protein (3MX5)

In the Biovia Discovery Studio tool²⁶, the 3MX5 protein was prepared for docking where it was first loaded in PDB format. The protein’s chain revealed where all the native ligands were attached. This was followed by the removal of all water molecules and heteroatom. Next, the addition of of polar hydrogen atoms was performed. The active site of the protein was defined using the active site bubble and the native ligands as guides. The coordinates were x = 17.7295, y = 26.7623, and z = 3.0223. From the prepared protein, incomplete charges were corrected and the resulting protein saved for further use²⁷.

Molecular docking of the ligands to the target protein

The ligands, one after the other, were docked against the LFHV nucleoprotein target protein, 3MX5. Molecular docking was performed as reported^27–31. First, the protein’s active site was identified with the aid of the Allosite online tool (https://mdl.shsmu.edu.cn/AST/). This was followed by the identification of the allosteric pocket using the molecular operating environment (MOE) site finder of the Molecular Operating Environment (MOE) version 2023.02, as previously reported²⁷. Following the identifications, the prepared protein and ligands utilised for molecular docking analysis as previously reported using the London dG scoring function²⁷. The docking scores and poses were visualised in 2 and 3D using the Biovia Discovery Studio and Pymol too, respectively.

Molecular dynamics simulation (MDS) studies

A 300 nanosecond (ns) MDS run for the studied ligands and the control co-crystal was carried out using the Schrodinger-Desmond module. A solvated system was created to ensure the presence of water molecules during the docking run, and this was achieved via the TIP3P model. A periodic boundary condition of 10 Å was utilised for the orthorhombic simulation box. Neutrality was achieved in the solvated system via the addition of adequate counter ions. In addition to the solvation and neutrality states, an isosmotic state was created and maintained in the simulation box by adding 0.15 M NaCl. Before the execution of the MDS, the system was allowed to reach equilibrium, and thereafter, at 310 K and 1.013 bar, a 300 ns MDS was carried out for the various complexes^16,27. In addition, the force field employed for the simulation was the OPLS (Optimized Potentials for Liquid Simulations). Furthermore, PCA and dynamic cross-correlation matrix analyses were also evaluated.

Density functional theory (DFT) evaluations

In addition to the MD and MDS carried out, all the ligands in this study were subjected to DFT evaluations. DFT has emerged as an invaluable quantum mechanical method that allows for the examination of electronic structure and the elucidation of potential reaction mechanisms³². The DFT computations were done using methods already reported using the Guassian 09 software³³. The computations were done using the basic sets of atoms, which were C, N, O, and H, and the B3LYP and exchange-correlation functional (ECF) with a 6–31 G (d, p). The geometry, configuration, and optimisation were done using the B3LYP/6–31 þ G (d, p) basis set to increase the reliability of the results^16,27.

Lipinski’s rule of five (LROF), ADMET and target protein predictions of the ligands

First, using the SWISSADME tool, the ligands were screened to evaluate their compliance with LROF and ADME properties at http://www.swissadme.ch/. The ADME properties were then validated using the ADMET tool of the pkCSM tool at https://biosig.lab.uq.edu.au/pkcsm/^34,35.

Results and discussion

Molecular docking

The docking output is presented in Table 1. All the ligands, including the control, returned higher docking scores than the control (co-crystal), apart from lauric acid, which returned a docking score of − 6.1 kcal/mol. The binding affinities for the rest of the ligands ranged from − 9.1 to − 12.3 kcal/mol. Compared to the standard drug, that is, sofosbuvir, all the test ligands, apart from lauric, gave higher scores that ranged from − 9.5 to − 12.3 kcal/mol. The test ligand with the highest score was NCTPMMDP with a score of 12.3 kcal/mol. The various docking scores indicate varying degrees of ligands and protein interactions. The high scores tentatively suggest that the test ligands bind more favourably to the target as much as the control and standard drug^37,38. All the test ligands in our study apart from lauric acid gave better binding affinities than those utilised in previous studies that ranged from − 6.0 to − 7.7 kal/mol [15 24}. Our findings align better with an earlier study that also docked various evodiamine derivatives against LFHV glycoprotein (5FT2) and nucleoprotein (3MX5)¹⁷. Their docking results in their results ranged from − 1.7 to − 8.5 kcal/mol and − 2.0 to − 11.0 kcal/mol, respectively against the glycoprotein and the nucleoprotein of LFHV in their study. Furthermore, the docking scores for the control drug for both proteins were − 5.9 and − 7.1 kcal/mol, respectively, for the control. In another study that also utilised evodiamine as ligands against Sars-Cov2 protein, the docking scores that ranged from − 6.31 to − 9.99 kcal/mol were observed²⁵. Compared to our study, the results indicate that our test ligands returned better interactions than those reported earlier, and this was also the case with the control drug. The observed differences could be as a result of the various amino acid residues involved in the different active sites of the proteins utilised in the various studies, and the various viruses involved in the various studies.

Table 1.

Docking affinities for various ligands.

Ligand	Binding energy (kcal/mol)
3mx5_ Control(co-crystal)	− 8.9
PubChem ID 45,375,808	− 9.1
PubChem ID 3893	− 6.1
PubChem ID 49,804,500	− 9.8
PubChem ID 49,804,912	− 12.3
PubChem ID 56,967,508	− 9.5

Key: PubChem ID 45,375,808 = Sofosbuvir; 3893 = Lauric acid; 49,804,500 = NCTPMMDP = N-[[2-(3-chlorophenyl)-5-(trifluoromethyl)pyrazol-3-yl]methyl]-2-(1-methyl-2,3-dihydroindol-5-yl) propanamide; 49,804,912 = N-(2-Fluorobenzoyl) evodiamine; 56,967,508 = 5-(Methylene) evodiamine.

Added to the binding scores or affinities, the docking poses or favourable docking positions of the ligands and the target, as well as the control drug and co-crystal, were visualised in 2D and 3D (Figs. 1 and 2) to reveal the various amino acids of the target protein that interacted with the ligands. Figure 1 shows the poses in 2D. From the result, it can be seen that the ligands interacted differently with the target protein. The co-crystal gave the most amino acids (n = 9) that interacted with the visible bonds depicted using arrows (see legend in Fig. 1). For the rest of the ligands depicted with labels B to F, the amino acids with visible bonds (in arrows) were 3, 6, 4, 3, and 2, respectively, for ligands B, C, D, E, and F. The main interaction types were greasy (Van der Waal, shown in green) and polar interactions (shown in pink).

Fig. 1.

Visualisation outputs of the various ligands co-crystal to the target protein in 2D. Key: Figures A to F shows the 2 poses for the various ligands, while the unlabeled portion shows the various bond types.

Fig. 2.

Visualisation outputs of the various ligands co-crystal to the target protein in 3D. Key: Figures A to F shows the 3D poses for the various ligands, while G and H show the various protein chains and ligand in the active site of the target protein.

Figure 2 reveals the binding poses interactions in 3D (A to F) and the structure of the target proteins (G and H) in various modes. It reveals the interacting amino acids, and these were for ligand A: Trp164, Arg323, Thr178, Lys253, Phe176, Ile241, Lys309, Arg300, and Asn249. For ligand B, the amino acids were Asn174 and Arg329, while for ligand C, the amino acids were Arg300, Asn305, Leu239, PHe176, and Arg323. For ligand D, the amino acids were Asn305, Phe176, Arg300, and Glu117, while for ligand E, the amino acids were Gln175, Asn174, Arg329, Phe176, Ile241, Asn240, Glu117, Try308, Trp164, and Arg300. For ligand F, the amino acids were Phe176, Leu172, and Try308. This implies that ligand F, 5-(methylene) evodiamine, had the highest number of interacting amino acid residues, one residue more than the co-crystal and eight (8) more than the control drug, sofosbuvir. The various amino acid residues observed, their receptors, and how they interacted with the various atoms of the ligands are summarised in Table 2. The result indicates that the co-crystal returned the highest number of interactions (n = 33), followed by sofosbuvir (n = 6), lauric acid (n = 3), NCTPMMDP and evodiamine (n = 4 each), and 5-(Methylene) evodiamine (n = 2). The dominant interactions were H-acceptors and ionic bonds. Hydrogen acceptors and donors indicate the presence of hydrogen bonds³⁹.

Table 2.

A summary of the various interacting atoms, their interactions, distances and binding energies.

Ligand	Receptor	Interaction	Distance	E (kcal/mol)
3mx5 / 3mx5_ Control(co-crystal)
O5’ 7	NH2 ARG 300	H-acceptor	3.04	− 1.2
O1A 22	NH1 ARG 300	H-acceptor	2.69	− 8.3
O1A 22	NZ LYS 309	H-acceptor	2.73	− 15.1
O1B 23	N THR 178	H-acceptor	2.72	− 6.3
O1B 23	OG1 THR 178	H-acceptor	2.62	− 2.8
O1G 24	NH1 ARG 300	H-acceptor	2.60	− 6.7
O2 26	N ASN 240	H-acceptor	2.77	− 3.7
O2B 28	NZ LYS 253	H-acceptor	2.76	− 2.1
O2B 28	NZ LYS 309	H-acceptor	2.72	− 11.7
O2B 28	NH1 ARG 323	H-acceptor	2.76	− 8.2
O2B 28	NH2 ARG 323	H-acceptor	2.75	− 8.5
O3G 34	NZ LYS 253	H-acceptor	2.64	− 19.4
O3G 34	ND2 ASN 305	H-acceptor	2.75	− 7.6
O3G 34	NZ LYS 309	H-acceptor	2.74	− 2.5
O4 36	NE1 TRP 164	H-acceptor	3.04	− 1.8
O1A 22	NH1 ARG 300	Ionic	2.69	− 6.9
O1A 22	NH2 ARG 300	Ionic	3.77	− 1.0
O1A 22	NZ LYS 309	Ionic	2.73	− 6.5
O1A 22	NH1 ARG 323	Ionic	2.98	− 4.6
O1B 23	NZ LYS 253	Ionic	3.89	− 0.7
O1B 23	NH2 ARG 323	Ionic	3.88	− 0.8
O1G 24	NE ARG 300	Ionic	3.55	− 1.7
O1G 24	NH1 ARG 300	Ionic	2.60	− 7.8
O1G 24	NZ LYS 309	Ionic	2.74	− 6.5
O2A 27	NH1 ARG 323	Ionic	2.98	− 4.6
O2A 27	NH2 ARG 323	Ionic	3.86	− 0.8
O2B 28	NZ LYS 253	Ionic	2.76	− 6.3
O2B 28	NZ LYS 309	Ionic	2.72	− 6.6
O2B 28	NH1 ARG 323	Ionic	2.76	− 6.3
O2B 28	NH2 ARG 323	Ionic	2.75	− 6.4
O3G 34	NZ LYS 253	Ionic	2.64	− 7.4
O3G 34	NZ LYS 309	Ionic	2.74	− 6.5
6-ring	6-ring PHE 176	pi-pi	3.48	− 0.0
3mx5 / 3893_model1_out
O2 36	N ASN 174	H-acceptor	3.09	− 1.6
O2 36	NH1 ARG 329	H-acceptor	2.81	− 2.3
C1 7	6-ring PHE 176	H-pi	3.95	− 0.8
3mx5 / 45375808_model1_out
C3 43	O SER 238	H-donor	3.33	− 0.7
O6 3	NH1 ARG 323	H-acceptor	3.07	− 3.3
O9 25	NZ LYS 253	H-acceptor	2.91	− 5.4
O9 25	NZ LYS 309	H-acceptor	2.90	− 8.7
O4 55	N ASN 240	H-acceptor	2.84	− 3.0
6-ring	CG GLU 117	pi-H	3.81	− 0.6
3mx5 / 49804500_model1_out
N2 11	OG SER 238	H-donor	2.87	− 1.7
N1 32	NZ LYS 309	H-acceptor	2.85	90.0
N3 1	OE2 GLU 117	Ionic	3.34	− 2.6
N4 2	OE2 GLU 117	Ionic	2.63	− 7.5
6-ring	6-ring PHE 176	pi-pi	3.81	− 0.0
3mx5 / 49804912_model1_out
O1 1	N ASN 240	H-acceptor	2.84	− 4.7
O2 2	NH2 ARG 300	H-acceptor	3.12	− 1.9
5-ring	NH1 ARG 300	pi-cation	3.64	− 1.6
6-ring	6-ring PHE 176	pi-pi	3.98	− 0.0
3mx5 / 56967508_model1_out
C3 8	OE2 GLU 117	H-donor	3.23	− 0.9
N2 3	NH2 ARG 300	H-acceptor	2.91	70.9

Molecular dynamics simulation (MDS) evalutions

Given the inability of docking as a technique to reveal the overall thermodynamics and stability of the complexes formed, the molecular dynamics simulation evaluations were performed as previously reported^16,27. For all the complexes formed, a 300 ns MDS run was performed^16,27. The MDS runs were performed for both the complexes and the individual ligands, and the results are presented in Figs. 3, 4, 5, 6 and 7. Figure 3 shows the result of the RMSD analysis for the various complexes formed in the study. The result indicates fluctuations throughout the simulation runs for all the complexes formed. For the control (co-crystal), the RMSD values showed fluctuations over the entire simulation period. Its RMSD values that were lower than 2Å were observed around 20, 100–150, 200–220, and at 270–300 ns. For ligand 3893, regions of stability for the ligand were observed for the first 130 ns and thereafter, values higher than 2Å were observed, indicating instability. For the complex formed between ligands 45,375,808 and 49,804,500 and the target protein, the RMSD values only showed minimal stability at the first 50 ns of the simulation, while for the rest of the simulation, they were above 2Å for both the ligand and the target protein except for ligand 49,804,500 that gave stable values at 100–150 and 260–300 ns periods. The ligand 49,804,912 gave the most stable values for the majority of the simulation period with values less than 2.5 Å, except for the first 30 ns that were above 3Å. For ligand 49,804,912, the first 0-150 and 250–300 ns gave values that were less than 2 Å indicating stability. Overall, the ligands returned more stable values than the protein.

Fig. 3.

RMSD values of the various complexes.

Fig. 4.

RMSF values of the formed complexes.

Fig. 5.

A graphical illustration of the contacts (protein-ligand).

Fig. 6.

Frequency of contacts made by the various amino acids during simulation.

Fig. 7.

Ligand interaction properties (RMSD, rGyr, MolSA, SASA and PSA).

In addition to the RMSD values, the RMSF was also evaluated for all the complexes formed (Fig. 4). The RMSF is a structural feature that is revealed when simulation is performed. It reveals changes in the amino acid residues during the simulation runs. For the control co-crystal, regions of high RMSF values above 2.4 Å were observed for amino residues at 130, 320, and 460, while for the complex with ligand 3893, high RMSF values higher than 2.4 Å were observed at residues 200, 320, and 460. For the complex between 45,375,808 and the target protein, spikes above 2.4Å were noted around amino acid residues 120 to 149, 320, and 460, while for 49,804,500 and the target, the spikes were observed around residues 150, 320, and 380–400 and 460. Similar high spikes were observed for the complexes formed between ligands 49,804,912 and 56,967,508 and the target protein, and these were at residues 150, 200, 320, 380, 400, and 460. Compared with the co-crystal control, the various complexes formed between the test ligands and the standard drug were all within the high spikes observed and were similar. Overall, in all the complexes, there were more regions with low RMSF than those with high RMSF, indicating that those regions with low RMSF values showed little divergence from the control co-crystal and the stability of the amino acids in those regions.

In addition to the RMSF evaluation, an evaluation of the amino acid residues in the various complexes was also carried out (Fig. 5). The result revealed the various amino acids that made contact with the ligands and the bonding type utilised in the process. The ligand (45375808)-protein complex contact returned 44 amino acids, while that of the control returned a total of 34 amino acids. The rest of the complexes returned 28, 52, 30, and 18 amino acid residues in contact, and these were complexes formed with ligands 56,967,508, 3893, 49,804,500, and 49,804,912, respectively. The top three bonding types in all the complexes were hydrogen, water bridges, and hydrophobic bonds. These bonding types were reported in an earlier study that also evaluated ligand-contact interaction^39,40. These bond types were also observed during the docking evaluations. Collectively, these bonds determine the binding affinities of the ligands to the target protein^41–43. Specifically, the presence of hydrogen bonds is known to provide specificity and stability via the aligning of the ligands within the vicinity of the protein’s binding pocket, while hydrophobic interactions are also pervasive and are known to enhance the binding strength by minimising desolvation energy of the formed complexes. On the other hand, salt bridges have been shown to promote electrostatic complementarity. Furthermore, as a type of bond, H bond play an integral role in ligand-protein interactions where they function as either an electron acceptor or donor, as seen earlier in the docking results⁴⁰. Ionic/polar interactions are the interactions of two opposite charges that are several distances apart but do not have a H bond. Water bridges, on the other hand, are H-driven protein-ligand interactions driven by a water molecule³⁹. The hydrogen bond is also important in drug discovery, specificity, metabolism, and adsorption³⁹. As observed in all the complexes, apart from the complex formed between ligand 49,804,912 and the target protein, the most common amino acid residue Arg300 made contact with the ligands using either with or ionic bond or a water bridge. Additionally, Leu120 consistently made contact with the various ligands even in the control co-crystal using only a hydrophobic bond, while Glu117 consistently utilised hydrogen and water bridges or ionic and water bridges. The presence of unique amino acids and their bonds is key to optimising small molecules as drug candidates⁴⁰.

In addition to the ligand-protein contact analysis, the actual number of contacts made during the simulation runs is depicted in Fig. 6. The top-level plot for each ligand-protein interaction shows the number of contacts made during the simulation run. For the various ligands and their respective complexes, the number of contacts made were 30, 9, 15, 9, 9, and 6 for the control, 3893, 45,375,808, 4,980,500, 49,804,912, and 56,967,508, respectively. Furthermore, the number of contacts made by the individual amino acids depicted by the brown colour legend to the left of each plot indicates that the control had more amino acids that made contacts with the respective ligands during the 300 ns simulation run. The result also indicates that for complexes formed with ligands 3893, 49,804,500, 4,537,808, and, to a lesser extent, 49,804,912, the co-crystal uniformly made contact throughout the 300 ns simulation run.

Ligand properties analysis

Using simulation, the properties of our ligands were further explained using five parameters: RMSD, radius of gyration (rGyr), molecular surface area (MolSA), solvent accessible surface area (SASA), and polar surface area (PSA). Figure 7 shows the outputs of the various ligand properties following 300 ns of simulation run. The RMSD values ranged from 0.8 to 1.5, 1–3.5, 1.5–3.5, 0.8–2.4, 0.4–1.6, and 0.25–0.75 Å, respectively, for the control and ligands 3893, 45,375,808, 49,804,500, 49,804,912 and 56,967,508, respectively. Generally, RMSD values less than 2 Å indicates stability, implying that ligands 49,804,912 and 49,804,500 were the most stable. Furthermore, the radius of gyration also reveals fluctuations. The radii range were 4.2–4.6, 3.0–5.4, 4.0–5.2, 4.5–5.4, 3.92–4.08, and 3.5–3.7 Ų, respectively, for the control and ligands 3893, 45,375,808, 49,804,500, 49,804,912 and 56,967,508. The radius of the various ligands, when compared to the control, did not vary so much, indicating an overall compactness of the complexes⁴⁴. The molSA indicates minor fluctuations for all the ligands as well as the control co-crystal. Steady values were reached at 330, 260, 450, 430, 380, and 300 Ų, for the control, and ligands 33,893, 45,375,808, 49,804,500, 49,804,912 and 56,967,508, respectively. Similarly, steady values were reached for the SASA values at 50, 50, 90, 90, 60, and 50 Å^2, respectively, for the control and ligands 3893, 45,375,808, 49,804,500, 49,804,912 and 56,967,508, while for the PSA, steady values were obtained at 460, 95, 180, 50, 48, and 20 Ų, respectively, for the control and ligands 3893, 45,375,808, 49,804,500, 49,804,912 and 56,967,508. The values of molSA, SASA, and PSA are measures of the surface area available for contact with water molecules. The larger the surface area, the more complexes are available for contact with water. As a parameter, it is useful in predicting the stability and folding of a molecule⁴⁵.

Principal component analysis (PCA) and dynamic cross-correlation matrix analysis

We further evaluated the nucleoprotein, 3MX5, dynamics using PCA as previously reported^43–45 and the results presented in Figs. 8 and 9. The evaluations were done using the generated eigenvalues and eigenvectors. Both variables are important in the description of key dynamics of a protein, such as motion, folding, and any conformational changes in its backbone^46–48. The motion of the formed complexes in our study varied (Fig. 8). Consistently, in all the complexes formed, the motion of the complexes was accounted for by principal components (PC) 1, 2, and 3. For the control, depicted as A, the values were 53.01, 8.34, and 6.21, while for B, the values were 36.23, 20.56, and 15.32. For ligands C and D, the values were 43.55, 10.25, and 8.17, 46.95, 8.37, and 6.18, respectively. For ligand D, the values were 40.81, 17.06, and 5.38, while for ligand E, their respective PC 1, 2, and 3 values were 28.01, 16.63, and 9.58. The variations in the PC values observed align with previous reports indicating conformational changes^43–45. The PC1 values indicate that the ligands E, D, and C had the highest values, indicating that the complexes they formed were subjected to the highest conformation changes after the control ligand. Figure 9 shows the result of the dynamics cross-correlation matrix for all the test ligands. The dynamics cross-correlation analysis allows for the evaluation of the positive and negative impacts of amino acids^46–48. The correlation values ranged from − 1 to + 1, that is, from strong negative to positive correlations (see colour legend in Fig. 9). Apart from ligand 56,967,508, all the ligands had more positive interacting amino acids, thus more cumulative positive interactions or correlations than repulsive.

Fig. 8.

PCA and eigenvalue rank plots for all the ligands. Key: Top right = control while for ligands 3893, 4,537,808, 49,804,500, 49,804,912 and 56,967,508, top left, middle left, middle right, bottom left and bottom right/.

Fig. 9.

Dynamic cross-correlation matrix for the ligands. Key: Top right = control while for ligands 3893, 4,537,808, 49,804,500, 49,804,912 and 56,967,508, top left, middle left, middle right, bottom left and bottom right/.

FMO analysis of the various ligands

The results of the Frontier Molecular Orbitals (FMO) analysis is presented in Fig. 10; Table 3. Analyzing the FMO yields significant insights into their chemical properties, potentially informing drug discovery efforts and therapeutic strategies against Lassa fever¹⁵. The energy levels of the FMO, as represented by the Highest Occupied Molecular (HOMO) and Lowest Unoccupied Molecular Orbital (LUMO), alongside the energy gap (ΔE), are crucial indicators of molecular reactivity¹⁵. A smaller ΔE suggests heightened reactivity, which could be exploited in targeting these nucleoprotein for therapeutic intervention¹⁵. Moreover, parameters such as electronegativity (χ Pauling), global electrophilicity index (ω eV), and chemical potential (µ) offer insights into the nucleoprotein’s electron donation, acceptance capabilities, and potential for work transfer, respectively¹⁴. These factor are pivotal in understanding the interactions of the nucleoprotein with other molecules, including potential drug candidates⁴⁹. Additionally, hardness (η) and global softness (σ) shed light on the stability and reactivity of the nucleoprotein, essential considerations in drug design and optimization⁵⁰. The values of the highlighted parameters are presented in Table 3. Therefore, in comparing the five proposed Lassa fever virus nucleoprotein based on their FMO parameters, significant variations.

Fig. 10.

HUMO-LUMO energy analysis of the various ligands.

Table 3.

FMO parameters of the various ligands.

Ligands	Lumo	Homo	ΔE in eV	χ Pauling	η in eV	σ	µ in eV	S	ω ev	Pubchem IDs
Lauric acid	0.06874535	− 0.2607218	0.329467175	− 0.09599	0.164734	6.070407	0.095988	3.035204	0.027966	3893
Sofosbuvir	0.031427633	− 0.215992422	0.247420055	− 0.09228	0.12371	8.083419	0.092282	4.04171	0.034419	45,375,808
N-[[2-(3-chlorophenyl)-5- (trifluoromethyl)pyrazol-3-yl]methyl]-2-(1- methyl-2,3-dihydroindol-5-yl) propanamide	− 0.100695902	-0.294780145	0.194084243	− 0.19774	0.097042	10.3048	0.197738	5.152402	0.201461	49,804,500
N-(2-Fluorobenzoyl) Evodiamine	0.06675338	− 0.199010451	0.265763831	− 0.06613	0.132882	7.525479	0.066129	3.762739	0.016454	49,804,912
5-(Methylene) Evodiamine	− 0.098762173	− 0.336175237	0.237413064	− 0.21747	0.118707	8.424136	0.217469	4.212068	0.1992	56,967,508

emerge. Lauric acid exhibits a relatively larger energy gap (ΔE = 0.329467175 eV) compared to others, suggesting lower reactivity, while compounds like NCTPMMDP (ΔE = 0.194084243 eV) and 5-(Methylene) evodiamine (ΔE = 0.237413064 eV), and N-(2-Fluorobenzoyl) evodiamine (ΔE = 0.265763831) display smaller energy gaps, indicative of higher reactivity.

Electronegativity values vary across the compounds, influencing their electron donation and acceptance capabilities, such as lauric acid (-0.09599), sofosbuvir (– 0.09228), and NCTPMMDP (– 0.19774), 5-(Methylene) evodiamine (– 0.21747), and N-(2-Fluorobenzoyl) evodiamine (-0.06613). Differences in hardness (η) and global softness (σ) reflect variations in stability and polarizability. For instance, NCTPMMDP exhibits the least hardness and highest softness (0.097042 eV and 10.3048, respectively) compared to lauric acid (0.164734 eV and 6.070407), sofosbuvir (0.12371 eV and 0.12371), N-(2-Fluorobenzoyl) evodiamine (0.132882 eV and 7.525479), and 5-(Methylene) evodiamine (0.118707 eV and 8.424136). The HOMO-LUMO iso-surface plot is provided in Fig. 10 to describe the electron-rich region (HOMO) and electron-deficient region (LUMO). N-[[2-(3-chlo rophenyl)-5-(trifluoromethyl)pyrazol-3-yl]methyl]-2-(1-methyl-2,3-dihydroindol-5-yl) propenamide is observed to possess higher reactivity as a result of a smaller energy gap, the smallest chemical hardness, and the highest softness, indicating increased reactivity compared to the other studied compounds. This is followed by 5-(Methylene) evodiamine and Sofosbuvir, which possess a relatively lowered energy gap, reduced chemical hardness, and high softness, indicating relatively high chemical reactivity.

The gradient observed among the ligands as evidenced by the docking scores was strongly influenced by the electronic properties of the ligands, such as electronegativity, energy gaps, chemical hardness, and softness⁵¹. In our study, the highest docking scores was attained by N-(2-Fluorobenzoyl) evodiamine. This can be attributed to its electronic properties properties that included moderate electronegativity, a relatively low energy gap, reduced chemical hardness, and high softness. These properties put together enabled the ligand to form strong and diverse interactions, such as hydrogen bonds, hydrophobic contacts, and van der Waals forces, within the binding pocket. The molecular dynamics simulations evaluation further supported the findings via the observed stable RMSD, RMSF, and rGyr values, as well as significant conformational changes in the protein structure observed through PCA analysis. This underscores the ligand’s capacity to establish and maintain robust protein-ligand interactions and aligns with the previous findings⁵¹. On the other hand, 5-(Methylene) evodiamine, while not achieving the highest docking score, still exhibited commendable activity that was driven by its very low energy gap and high softness. These properties of the ligand may have enhanced its adaptability and to the binding pocket of the target protein, allowing it to engage in stable interactions with the target protein. Its RMSD and PCA results further demonstrated its ability to induce conformational changes in the protein, reinforcing its potential as a viable inhibitor. Compared to N-(2-Fluorobenzoyl) Evodiamine and 5-(Methylene) evodiamine, sofosbuvir showed low energy gap and high softness, supporting its ability to form stable hydrogen bonds and polar interactions, Collectively, these findings highlight the importance of electronic properties in predicting docking outcomes. Specifically, our study indicates that ligands with lower energy gaps and higher softness are more reactive and adaptable, enabling stronger and more stable protein-ligand interactions.

ADMET prediction

The process involved in the discovery of a new drug is complex. However, it is evolving rapidly with the advent of tools to predict ADMET properties, an important set of parameters in the drug development process⁵². The ADMET properties of the various ligands and the co-crystal were predicted to further access their drug-like ability. First, the molecular descriptors were evaluated to reveal compliance with the LROF. The LROF advocates that for a compound to be regarded as having the potential to be orally absorbed effectively, it has to violate none or just one of the rules³⁴. These rules include molecular weight (MW ≤ 500), octanol/water partition coefficient (AlogP ≤ 5), hydrogen bond donor-HBD (≤ 5), and hydrogen bond acceptor (≤ 10). Only sofosbuvir violated one of the rules (MW > 500). Violation of one of the rules does not make the drug less orally active^34,52. This implies that all the ligands and the control drug would be active orally. Next, to further evaluate their drug-likeness, their ADMET properties were evaluated.

The result of the molecular descriptors is presented in supplementary Table 1, while the result of the pharmacokinetic properties (ADMET) of various ligands was predicted using the pkcsm tool and is presented in supplementary Table 2. The adsorption parameters examined were water solubility, CaCO₂ permeability, intestinal absorption (human), skin permeability, P-glycoprotein substrate, P-glycoprotein I inhibitor, and P-glycoprotein II inhibitor. The water solubility values of the ligands ranged from − 5.467 to – 4.181, while that of the control standard drug was − 3.953. The values overall indicate the solubility of the compounds⁵³. The CaCO₂ permeability values of the various ligands were all higher than the ideal value of > 0.90⁴⁹. The intestinal absorption values for the various ligands (> 90%) indicate that all of them, including the standard drug with the least value of 64.31, were all above the poor absorption cut-off value of 30%. This implies that the ligands have the capacity to survive the various intestinal hurdles and still have a good tissue distribution⁵³. The values for skin permeability being greater than log Kp > − 2.5 for all the ligands depict poor skin permeability³⁵. The VDss values for ligand 5-(Methylene) evodiamine was the highest, followed by NCTPMMDP.

Their high VD values indicate that they have the potential to reach tissues and not just plasma³⁵. BBB permeability being greater than − 0.3 for a given drug indicates poor brain distribution, while ligands 3893 and 45,375,808 showed the ability to cross the CNS with values greater than − 2³⁵. Ligands 3893 and 45,375,808 were not substrates or inhibitors of the cytochrome system, an important detoxifying enzyme system in the body. Other ligands showed variable results in the cytochrome system. The excretion results for the various ligands indicate variable results for renal OCT2 substrate and total clearance. Only the ligand 56,967,508 was a substrate for renal OCT2. Toxicity prediction indicates that only ligands 56,967,508 and 49,804,500 showed cancer-causing potential as they were positive to the AMES toxicity test, while the standard drug, sofosbuvir, N-[[2-(3-chlorophenyl)-5-(trifluoromethyl)pyrazol-3-yl]methyl]-2-(methyl-2,3-dihydroindol-5-yl) propanamide and N-(2-fluorobenzoyl) evodiamine showed potential to be hepatotoxic³⁵. The hepatotoxicity concerns for both compounds can be handled by modification of both ligands to remove potentially toxic function groups or via the use of encapsulation and controlled release⁵⁴.

Chemically, the ligands used in the study possess diverse chemical structures. 5-(Methylene) evodiamine and N-(2-fluorobenzoyl) evodiamine are derivatives of evodiamine. The former possess a methylene group and an alkaloid ring, while the latter possess a fluorobenzoyl moiety attached to the evodiamine core. On the other hand, NCTPMMDP possesses a complex structure that contains the pyrazole-indole derivative that contains multiple functional groups, which include the trifluoromethyl group and a chlorophenyl ring. Lauric acid possesses carboxylic acid as its main functional group. The alkaloid ring, which is a key structure of evodiamine, is known for its diverse biological roles^55–57, even against SARS-CoV-2 main protease and spike glycoprotein⁵⁸. Our docking result for the evodiamine derivatives indicates that they can interact with key amino acid residues within the milieu of the active site of the target protein via the observed hydrogen bonds and hydrophobic interactions, stabilising the protein-ligand complexes formed. The observed strong docking exhibited by these derivatives returned strong docking scores, indicating a strong binding affinity to the 3MX5 protein aligns with an earlier report that indicated that evodiamine derivatives can bind to both glycoprotein and nucleoprotein of lassa fever virus¹⁷. These compounds most probably inhibit the viral protein by stabilising its inactive conformation, thereby blocking essential viral replication functions. On the other hand, NCTPMMDP possession of diverse groups enhances its electron-withdrawing and lipophilic properties. Thus, it is probable from our docking studies that this compound is capable of forming stable hydrogen bonds and van der Waals interactions with key residues that are central to the activity of the target protein and, in the process, block its enzymatic activity. The simple structure of lauric acid was visible by the lowest binding energy it returned via fewer interactions. Lauric acid as a ligand in this study may exert its anti-viral activity by inserting itself into the lipid layer of the virus and in the process disrupt the membrane activity of the virus⁵⁸. Sofosbuvir, unlike the rest of the ligands, is an approved drug; it is a nucleotide analogue pro-drug that targets RNA-dependent RNA polymerase (RdRp)⁶⁰. Even though our target protein is a nucleoprotein, it is essential for encapsidating the genome of the virus and also enables the activity of the RNA-dependent RNA polymerase (RdRp) activity⁶¹. The present study hold a lot of prospects for the development of specific therapeutics against LFHV that is still elusive at the moment. In this study we have identified some ligands (n = 5) including the control drug with potential anti LHFV properties and this provides the theoretical basis for advancing the course for an anti-LFHV therapeutic. Furthermore, the findings in this offer a reproducible methodology to guide the screening of future compounds.

Conclusion

The aim of the study was to evaluate the anti-LHFV properties of -(Methylene) evodiamine, N-(2-Fluorobenzoyl) evodiamine, N-[[2-(3-chlorophenyl)-5-(trifluoromethyl)pyrazol-3-yl]methyl]-2-(1-methyl-2,3-dihydroindol-5-yl) propanamide, lauric acid, and sofosbuvir against the target protein, 3MX5, using entirely an in silico study approach. Docking scores for the test ligands ranged from − 9.5 to -12.3 kcal/mol, while that of the control and the standard drug were much lower and ranged from − 8.9 to -9.1 kcal/mol. Overall, the scores indicate a favourable binding to the active site of the target protein by the test ligands. The RMSD for the test ligands as well of the control all returned stable values that were generally less than 2.5 Ų. Similar low RMSF values (less than 2.4 Ų.) were also observed for the ligands and the control except for amino acid residues 150, 200, 320, 380, 400, and 460, respectively that recorded higher values. Put together, the RMSD and RMSF values of the ligands were indicative of ligand-protein complexes that were stable at some intervals of the simulation runs. The ligand interactions studies revealed stable ligand, as revealed by the favourable rGyr and RMSD values. The molSA, SASA, and PSA values corroborated the predicted absorption and Lipinski findings. Except for the control that had a MW greater than 500 g/mol, the test ligands all obeyed the Lipinski rule. All the ligands drove strong conformational changes within the backbone of the protein except for lauric that gave the least PCA values. Except for ligand 56,967,508, other ligands returned an excess of positive interacting amino acids that cumulated into more positive interactions than repulsive. Overall, the various test ligands showed more favourable ADMET and docking properties than the control and standard drug. Future studies should include in-vivo and in-vitro experimental studies, exploration of a broader number of ligands and viral targets.

Limitations of the study

The present study provided excellent insights into the potential anti-LHFV properties of the tested ligands; however, there were some limitations. First, the study was an in silico study in design. Second, our simulation time frame was kept at 300 ns. Third, our tested ligands were small in terms of number. Future studies should focus on in vivo and in vitro designs, while future in silico studies aimed at assessing the anti-LHFV properties of the tested ligands or more ligands should utilise longer simulation times. Despite the limitations, the present study offers a number of advantages. First the study leverages on silico approaches and has identified promising lead compounds with better ADMET properties. Furthermore, our findings had provided useful insight into the interactions of the test ligands and the target protein, stability dynamics, and its changes of their complexes.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

AEE, UOE, EM, IUB and MA conceived the project/idea; Funding: AEE, UOE, EM, IUB and FON; Methods: UOE and MA; Analyses: MA, SAM, EM, JP, SA, SA, DEC, and UOE; Writing of draft: UOE, EM, IUB, DEC, MA and JP. Review: All the authors reviewed the manuscript.

Funding

The study was funded by the 2021 Tertiary Education Trust Fund (TetFund) National Research Fund (NRF) (NRF2021_SETI_HSW_00478).

Data availability

Data is provided within the manuscript or supplementary information files.

Competing interests

The authors declare no competing interests.